[tor-reports] Karsten's March

[Karsten Loesing]

Here are the top 5 five things I wanted to do in March:

> - Fix too high directory fetch retry schedules (#6752).

Done.  Identified and fixed remaining bootstrapping issues in testing
networks from too high directory fetch retry schedules and wrote a
branch to be reviewed and merged into 0.2.5.x.

> - Finish the Torperf rewrite in Twisted, this time for real.

Half done.  Made some good progress on the Twisted rewrite of Torperf
[0] based on work by meejah, peer, atagar, and ln5.  The current status
is that it makes requests over Tor to its own webserver and captures all
relevant stream and circuit information.  However, despite being a lot
of work (of course, much more than expected), this is all just a
prerequisite for the actual deliverable, which is making requests to
real websites using Selenium/Firefox.  Will have to continue working on
this deliverable in April and maybe even in May.  Twisted is, well, twisted.

> - Wrap up sponsor F year 3 phase 1.

Done.  We completed six phase 1 items and deferred three of them to
phase 2, one of them being in the must-do section.  See the sponsor F
year 3 wiki page [1] for details.

> - Kick off sponsor F year 3 phase 2.

Half done.  Discussed first steps for many phase 2 deliverables, but not
all of them.

> - Attend dev meeting in Boston.

Done.  Had plenty of good discussions and generally a good time!


In addition to the tasks above, I worked on the following things:

- Changed the graphing engine in Atlas from flot to D3.js.  This was an
important step for #6361 ("Merge six bandwidth history graphs into
one"), and it was a good exercise for switching to D3.js in metrics-web
(or visionion).

- Added a new last_changed_address field to Onionoo that can be used to
ship a list of fallback directory mirrors on long-term fixed IPv6
addresses (#8374).


And here are the top 3 five things I want to do in April:

- Hand over sponsor F year 3 phase 2 coordination to Tom.

- Finish the Torperf rewrite in Twisted, this time for real, really.

- Start looking into implementing the new bridge user counting algorithm
(#8462) and into improving the direct user counting algorithm.


[0]
https://gitweb.torproject.org/karsten/torperf.git/blob/refs/heads/perfd:/README.perfd

[1]
https://trac.torproject.org/projects/tor/wiki/org/sponsors/SponsorF/Year3